The oracle of Delphi in the Temple of Apollo in Greece is one of the famous tourist attractions in the world. The oracle is dated from the 7th century BC. Before all the predictive analytics consulting and companies, the oracle is probably the very first and possibly the only ‘Prediction Engine’ that has worked effectively for 1100 years and the longer you at the temple, the more parallel with ML came to my mind.
How does the Ancient “Predictive Engine” work?
The ritual – Counselors traveled to the oracle and first have to provide information about the content including the nature of their questions, just like the modern data cleansing. Consequently, the oracle gifts were delivered accurately. Possibly, the first version of results as a service that pays for each request. The queries were directed to Pythia – priestess who sat on a tripod inhaling fumes right from a column in the floor of the temple.
The hazes were sent to the priestess into another state of consciousness so that she could use supernatural powers to find the answers to the questions of the counselors.
“Every prediction is uncertain, whether it’s on the basis of rules from machine learning models, dedicated systems, or the oracle of Delphi.”
The Uncertainty of Forecasts
Whether it’s about profit gain, longevity, or anything else, every prediction is uncertain to some extent. The data scientists quantify the vague by calculating confidence periods around the expected value. For this, the scientists use the changeability of the known distribution or results of Monte Carlo replications. With these periods, we can evaluate how broad the spectrum is in which the value can lie.
Dealing with Uncertainty
The oracle of Delphi is known for avoiding false predictions. For example, Croesus, King of Lydia, once was keen to visit the Oracle to see whether he should seek an alliance or attack the Persian Empire. And in answer, he received that the attack isn’t a good idea as it would destruct the entire kingdom. Croesus didn’t understand the ambiguity of the prediction and went into a battle and was defeated by Cyrus – the Persian ruler – the kingdom was his own. That’s how the Oracle was right eventually!
Let’s take another example of today’s life: “the marketing campaign will have sales figures.” A failure can be just as surprising as a sales hit – the prediction would be accurate in both cases. Of course, we can’t do our businesses based on this principle. But, blur in forecast isn’t something new and isn’t a side effect of machine learning. Uncertainties have always lived; only dealing with them was the different.
Transfer of Forecast in Business World
The prophecy of Pythia was not in human language while in her supernatural state of consciousness. It was just sounds and the other priests had to translate these sounds into meaningful dactylic hexameters.
In the same way, predictions generated by machines and software applications must also be translated into human language so that business users can understand it and implement it. This is a crucial step when it comes to decision-making.
With increasing awareness of Delphi, one Pythia was not sufficient to handle too many requests. So three Pythias were placed above the column where the fog escaped so that multiple requests can be handled and it divided the Prediction Engine workload – the first parallel processing in the history.
Predictions and humans have been working together to make significant changes in the world for centuries. However, it’s not certain what will happen in the future, but we can take early actions based on predictive analytics consulting services. Contact our team at ExistBI for further details.